A

Mean depth 1000 more than100×. regionsweresequencedatadepth analyzed attheindicateddepths;98.8%oftargeted B. Coveragesofthetargetedsequences sequencing on10PBLcases(meandepth=677). Figure S1. 200 400 600 800 0

PBL001_P

Depths and coverages of targeted deep sequencing.A.Depthsoftargeteddeep Depthsandcoveragesoftargeteddeep PBL002_T

PBL003_T

PBL004_T

PBL005_T

PBL006_T

PBL007_T

PBL008_T

PBL009_T

PBL010_T B

Mean coverage 100% 10% 20% 30% 40% 50% 60% 70% 80% 90% 0%

PBL001_P

PBL002_T

PBL003_T

PBL004_T

PBL005_T

PBL006_T ×20

PBL007_T

PBL008_T ×50

PBL009_T ×100 PBL010_T A B ×2 ×10 ×20 250 100% 200 90% 150 80% 100 Mean depth 50 70%

0 60%

50% PBL002_T PBL004_T PBL007_T PBL008_T PBL010_T PBL001_P PBL001_N PBL002_N PBL004_N PBL007_N PBL008_N PBL010_N PBL001_M1 PBL001_M2 PBL001_M3 PBL001_M4 PBL001_M5 PBL001_M6

Mean coverage 40%

30%

20%

10%

0% PBL002_T PBL004_T PBL007_T PBL008_T PBL010_T PBL001_P PBL001_N PBL002_N PBL004_N PBL007_N PBL008_N PBL010_N PBL001_M1 PBL001_M2 PBL001_M3 PBL001_M4 PBL001_M5 PBL001_M6

C D 14 80 12 70 10 60 50 8 40 6 30 4

Number of mutations Number of mutations 20 2 10 0 0 PBL002_T PBL004_T PBL007_T PBL008_T PBL010_T PBL001_P PBL001_P PBL001_M1 PBL001_M2 PBL001_M3 PBL001_M4 PBL001_M5 PBL001_M6

Nonsynonymous SNV Stopgain SNV Nonframeshift indels Frameshift indels Splice site

Figure S2. Sequencing coverages and numbers of mutations detected by WES. A. Depths of WES on tumor and normal DNA samples from 6 cases (mean depth = 124 for tumor samples and 141 for normal samples). B. Coverages of the exome sequences; 97.5% of the exome sequences were analyzed at a depth more than 20×. C-D. The number of somatic non-silent mutations and indels detected by WES and validated by PCR-based amplicon deep sequencing in 6 primary tumor samples (C) and 7 multiple samples from PBL001 (D). SNV, single-nucleotide variant. P M1 M2 M3 CLEC19A NOTCH1 SIK3 MFSD2B CTNNB1 UBE2E2 GOLM1 PTPRU CTSW KIAA0368 UBE2E2 UBE2E2 PTPRU CCDC88C CTNNB1 KDM6A CTNNB1 MRGPRE BRS3 CCDC88C SYNJ1 OR4D11 KDM6A OR4D11 CCDC88C DSP KDM6A MRGPRE PLXNB1 CTNNB1 CCDC88C SIK3 ZC3H18 OR4D11 ABCA13 PGK2_c.A155T NTNG1_c.A300T HUWE1 OR4D11 LEF1 MAGEB1 PGK2_c.G154A KLKB1 NTNG1_c.C301T LRRIQ3 TMEM132C_c.A2550G UBE2E2 PHKA2 EXPH5 ADGRG2 MUC3A ADCY1 BRWD1 ZBED2 FOXP4 MRGPRE LCN9 TMPRSS5 SIK3 MYBPC1 IQCF1 GRM2 CYP2W1 TMEM132C_c.G2551T MRGPRE ABHD12B KDM6A SERPINB7 FRMD6 HECW1 PTPRU HOXB13 LOC650293 ZNF804B MYO6 XCR1 SIK3 GLRA3 MADD KMT2A CFAP43 NARFL DNAH17 EDN2 CUX1 PCLO HIST1H4D NUDCD3 PTPRU SLC5A10 NLRP10 PCNXL3 DLL1 AOC1 CNBP IFT43 GBF1 CEP295 GSX2 MKI67 CADPS2 ASCC3 KRTAP13-2 GALR1 CELSR1 DMBT1 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 Cellular Frequency Cellular Frequency Cellular Frequency Cellular Frequency

M4 M5 M6

UBE2U SIK3 PKP3 KAT6B PTPRU MVK PIK3CD OGN CBS UBE2E2 OR51T1 KDM6A CCDC88C ADGB SIK3 KCNK17 ZBTB38 PPL MRGPRE BTBD16 MGMT OGN CTNNB1 OR51M1 MRGPRE CCDC88C MRGPRE GPR149 CTNNB1 OR51M1 FANCG OGN KDM6A OR51B4 SYNE1 KCTD1 UBE2E2 GLI3 PTPRU OR4D11 MLLT10 NPSR1 ARHGAP29 OR4D11 CUTC CDH3 PABPN1L UBE2E2 AHNAK OR4D11 DGCR14 TNXB OR51T1 ABCD4 SIK3 KDM6A GART PTPRU SETD4 ATR TIGD3 RAB11FIP3_c.C379A ZFHX3 OR51T1 CUL9 OR51M1 STIM1 CDKN2AIP GBP5 CTNNB1 XKRX ZNF516 RAB11FIP3_c.C380A TRNT1 KIAA2026 ADGRG4 RNF40 TRH HNRNPCL2 FA2H NTAN1 SLC17A6 MOXD1 SEMA3E CYSRT1 ZCCHC11 CD109 RPS6KA5 KCNQ5 COL6A5 RBM25 DMBT1 RANBP10 ALG1L TRH DLK1 LAMC3 KIAA1462 SLC34A2 TAF7L RNF38 HECW2 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 Cellular Frequency Cellular Frequency Cellular Frequency

Figure S3. Tumor subclonal populations in multiple samples from PBL001 estimated by PyClone. Cellular frequencies of detected somatic mutaions in each sample from PBL001 are shown on each panel. Mutations on chromosomal regions with copy number gains were not used for the analysis because mutant cell fractions may be underestimated. P, primary sample; M1-M6, metastatic samples. A

Primary PBL C>A C>G C>T T>A T>C T>G 0.15 0.10

0.05 Probability 0.00 TTT TTT TTT ATT TTA ATT TTA ATT TTA TCT TCT TCT CTT ATA TTC CTT ATA TTC CTT ATA TTC ACT ACT ACT TCA TCA ATC TCA CTA GTT TTG ATC CTA GTT TTG ATC CTA GTT TTG CCT CCT CCT TCC TCC ACA ACA TCC ACA ATG CTC GTA ATG CTC GTA ATG CTC GTA GCT GCT ACC CCA ACC CCA GCT TCG TCG ACC CCA TCG CTG GTC CTG GTC CTG GTC CCC CCC CCC ACG GCA ACG GCA ACG GCA GTG GTG GTG CCG GCC CCG GCC CCG GCC GCG GCG GCG

Metastatic PBL C>A C>G C>T T>A T>C T>G 0.15 0.10 0.05 Probability 0.00 TTT TTT TTT ATT TTA ATT TTA ATT TTA TCT TCT TCT ATA CTT TTC ATA CTT TTC ATA CTT TTC ACT TCA ACT TCA ACT TCA ATC CTA GTT TTG ATC CTA GTT TTG ATC CTA GTT TTG CCT ACA TCC CCT ACA TCC CCT ACA TCC ATG CTC GTA ATG CTC GTA ATG CTC GTA ACC CCA GCT TCG ACC CCA GCT TCG ACC CCA GCT TCG CTG GTC CTG GTC CTG GTC CCC ACG GCA CCC ACG GCA CCC ACG GCA GTG GTG GTG CCG GCC CCG GCC CCG GCC GCG GCG GCG

B Primary PBL Metastatic PBL

Strand directions Strand directions T T + − + −

A C G T C C G A G C G T C T C C G T

Figure S4. Mutational signatures of primary and metastatic PBL. A. Mutational signatures of 6 primary PBL samples in which WES was performed (upper plot), and 6 metastatic samples from PBL001 (lower plot). The probability bars are colored according to the 6 types of substitutions. Each substitution type contains 16 bars representing the sequence context immediately 3’ and 5’ to the mutated base. B. Mutational signatures consisting of the substitution patterns, the ±2 flanking bases, and the transcription strand directions, plotted for primary (left), and metastatic PBL (right). 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 2122

PBL001_P PBL001_M1 PBL001_M2 PBL001_M3 PBL001_M4 PBL001_M5 PBL001_M6 PBL002_T PBL004_T PBL007_T PBL008_T PBL010_T

CN

Loss Gain

Figure S5. SNP array-based copy number analysis of PBL. Copy number alterations in multiple samples from PBL001 and 5 primary tumor samples identified by SNP array analysis are shown. CN, copy number. A PBL001 PBL002

Chr11 Chr11

Hetero Hetero SNPs SNPs 1 1 0.5 0.5 ratio ratio Allelic Allelic 0 0

PBL003 PBL004

Chr11 Chr11 Hetero Hetero SNPs SNPs 1 1

ratio 0.5 0.5 ratio Allelic Allelic 0 0

PBL005 PBL006 Chr11 Chr11

Hetero Hetero SNPs SNPs 1 1

ratio 0.5

ratio 0.5 Allelic Allelic 0 0

PBL007 PBL008 Chr11 Chr11 Hetero Hetero SNPs SNPs 1 1 0.5 0.5 ratio ratio Allelic Allelic 0 0 B PBL010 PBL009 Chr11 Chr11

Hetero Hetero SNPs SNPs 1 1 0.5 0.5 ratio ratio Allelic Allelic 0 0

Figure S6. Targeted deep sequencing based allele-specific copy number analysis of chromosome 11. A. Sequencing-based copy number analysis identified CN-LOH with the minimal common region identical to chromosome 11p15.5 locus in 8 cases. Allelic ratios were calculated by allele frequencies and sequenced depths of SNPs; red/green dots represent major/minor allele frequencies at SNP sites. B. Allelic ratio of PBL009, the case with gain of whole chromosome 11, is also plotted. A

PBL002 (tumor purity = 0.57) PBL004 (tumor purity = 0.86)

TTCGTTCGTGGAAACGTTTC GGGTTATTTAA GTTACGCGTCGT TTC GTTC GTGGAAACGT TTCGGGTTAT TTAA GTTACGCGTCGT (T) (T) (T) (T) (T) (T) (T)

PBL010 (tumor purity = 0.44)

TTCGTTCGTGGAAACGTTTCGGGTTATTTAA GTTACGCGTCGT (T) (T) (T) (T) (T) (T) (T)

B PBL009 (tumor purity = 0.69)

TTCGTTCGTGGAAACGTTTCGGGTTAT TTAA GTTACGCGTCGT (T) (T) (T) (T) (T) (T) (T)

Figure S7. Bisulfite Sanger sequencing of ICR1 on chromosome 11p15.5. A. Bisulfite Sanger sequencing of 3 CN-LOH (+) cases (CpG sites are indicated by arrows). From the estimated tumor purity (see Methods), all samples examined were considered homozygously methylated, indicating that their CN-LOH causes paternal uniparental disomy. B. Bisulfite sequencing of PBL009, the case with gain of whole chromosome 11. A B * M 3.0 IGF2 ACTB ** * * 2.5

1000 bp 500 bp 2.0

100 bp 1.5 xpression of IGF2 e v 1.0 Relati PBL001 PBL005PBL009 PBL001 PBL005PBL009 0.5

Normal pancreas (Negative control) Normal pancreas (Negative control) 0 Normal PBL001 PBL005 PBL009 pancreas Sample

Figure S8. Relative expression levels of IGF2 in CN-LOH-negative samples. A. Reverse transcription PCR was performed in normal pancreas (as a normal control), PBL001 (as a CN-LOH-positive control), PBL005, and PBL009. ACTB (encoding β-actin) was used as a reference gene. The lanes of negative control contain water control. M, DNA size markers. B. DNA bands were quantified using Image Lab software (Bio-Rad). Relative intensities of IGF2 compared to ACTB were computed, and mean intensities ± s.d. of 3 technical replicates are plotted with corresponding P-values (Student’s t-test). *P < 0.05, **P < 0.01. A

3

2

1

Log2 ratio 0 CN = 2

-1

-2 1 0.5 ratio Allelic 0

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19202122 Chromosome

B C 3

2

1

0 Log2 ratio -1 -2

● 1 ● ●

● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● 0.5 ● ●● ● ratio

Allelic ● ● ● 0

Chr5 APC

Figure S9. Aberrant activation of Wnt signaling pathway in PBL009. A. Sequencing-based copy number analysis identified CN-LOH within chromosome 5q involving APC gene locus. Upper panel shows the genome-wide copy number profile in PBL009. A log2 ratio of zero (horizontal red line) corresponds to a normal, diploid copy number. A red arrow indicates the focal CN-LOH on chromosome 5q. Lower panel shows genome-wide allelic ratios. CN, copy number. B. Copy number changes and allelic ratios within chromosome 5. APC gene locus is designated on the idiogram with a red bar. C. Immunohistochemistry of β-catenin showing aberrant nuclear accumulation of β-catenin in PBL009. A

6 3

4 2 PBL Normal pancreas Density Density 1 2

0 0 0.00 0.25 0.50 0.75 1.00 0.00 0.25 0.50 0.75 1.00 β-value β-value B TSS TES 1.00

0.75 ylation 0.50 PBL Met h Normal pancreas 0.25

0.00 0.0 0.5 1.0 Relative position

Figure S10. Global DNA methylation profile of PBL. A. Distributions of β-value for each probe (left panel) and CpG island (right panel) are plotted. A higher unmethylated peak (β = 0, red arrow) and a lower methylated peak (β = 1, blue arrow) indicated the global hypomethylation in PBL. B. An average methylation on coding sequences were computed and plotted on the basis of relative position in each gene so that position 0.0 represents a transcription start site (TSS) and position 1.0 represents a transcription end site (TES). The whole coding regions were hypomethylated in PBL.